The Effect of Type of Paragraph on Native and Non-native English Speakers ' Use of Grammatical Cohesive Devices in Writing and Raters ' Evaluation

The present study examined the use of grammatical cohesive features in two types of paragraphs (chronologyvs. cause-effect) written by EFL learners and native speakers. In addition, the study investigated if the use of grammatical cohesive devices affected raters' evaluation of the paragraphs written by the EFL learners. To investigate the non-native speakers’ performances, 60 paragraphs) 30 cause-effect and 30 chronology paragraphs) written by 30 intermediate Iranian EFL learners were examined. Furthermore, to investigate those of native speakers, 20 paragraphs (10 cause-effect and 10 chronology paragraphs) were randomly selected from among the sample paragraphs presented in books teaching paragraph writing written by English native speakers. To analyze the data, Halliday and Hasan's (1976) taxonomy of cohesive devices was used, and MANOVA was conducted. To investigate raters’ judgments, four raters were asked to rate the paragraphs written by six randomly selected learners. The results showed that EFL learners more frequently used reference, conjunction, ellipsis and substitution, respectively in their chronology and cause-effect paragraphs. However, in their paragraphs of chronology, native speakers more frequently used reference, conjunction, ellipsis and substitution, respectively whereas in their paragraphs of cause-effect, they frequently used reference, conjunction, substitution, and ellipsis, respectively. A comparison of the performance of native speakers and non-native speakers revealed that native speakers used more reference in their paragraphs than non-natives did. As for raters’ judgments, the findings indicated that cohesion in the paragraphs was generally ignored by some raters.


INTRODUCTION
Writing can be a major criterion for better academic status and more educational success.It is the primary tool to communicate knowledge and present ideas (Hosseini et. al. 2013).However, writing may be a difficult task for learners.This difficulty stems from generating ideas, organizing the generated ideas and then translating them into areadable text.Therefore, writers must focus on planning and organizing (as higher level skills) and spelling, punctuation, word choice, etc. (as lower level skills) (Richards & Renandya 2002).
As one-of the textual elements of high quality writing, cohesion helps a text to be connected.In fact, cohesion refers to the explicit links in a text (Todd et. al. 2007).Halliday and Hasan (1976) maintain that cohesion helps in distinguishing a text from non-text and it is achieved through the use of cohesive ties, which connect different parts of a text.According to Halliday and Hasan, cohesive ties include reference, substitution, ellipsis, conjunction, and lexical relationships.Reference can be classified into pronominal, demonstrative, definite article, and comparative.Substitution consists of nominal, verbal and clausal ones.Ellipsis includes nominal, verbal and clausal ones.Conjunction is classified into additive, adversative, causative and temporal.And lastly, lexical cohesive devices include synonyms, antonyms and hyponyms.
The present study investigates an unsearched area, specifically how English native speakers and EFL non-native speakers use the patterns of grammatical cohesion in writing paragraphs of chronology and cause-effect.To do so, at first, the study makes a comparison between the proportion of grammatical cohesive devices used in paragraphs of chronology and paragraphs of cause-effect written by EFL learners.Next, the proportion of grammatical cohesive devices in paragraphs of chronology written by EFL learners are compared with that of native speakers.Then, a similar comparison is made with respect to the paragraphs of cause-effect.Furthermore, the study explores whether raters notice cohesion in evaluating EFL learners' writing samples of chronology and cause-effect.Therefore, the following research questions are addressed in the present study: 1. What is the effect of type of paragraph (chronology vs. cause-effect) on the proportions of grammatical cohesive devices used in Iranian EFL learners' writing samples and those of native speakers?2. Is raters' evaluation of paragraphs written by EFL learners affected by the cohesive devices used by the writers?How?
The findings of the study may, among other things, yield a model of how native speakers use grammatical cohesive devices which can be useful in teaching learners how to make textuality in the given types of paragraphs, namely paragraphs of chronology and paragraphs of cause-effect.Therefore, the present study can raise learners' consciousness of the importance and use of specific types of grammatical cohesive devices in each of these types of paragraphs.

LITERATURE REVIEW
A genre is generally defined as a socially recognized communicative event, containing some segments or moves, which serve a part of the text's communicative purpose.Each genre has its own structural patterns and quality (Swales 2004).The linguistic approach to study genres is a type of discourse analysis, which aims at identifying the specific nature of genres.It focuses on specific words and syntactic structures, which occur in both extremes (i.e., frequently or infrequently) in texts (Bhatia 1993).
In fact, texts are classified into five basic types, namely narration, description, argumentation, exposition, and instruction (Helder 2011, Sayah & Fatemi 2013).Narratives can bring access to the narrator's personal beliefs, emotions, attitudes and the socio-cultural norms they develop in their narrations.Description intends to represent a mental picture of a scene with objects, people and situations as perceived in space.Argumentative texts focus on the evaluation of relations between concepts, the expression of an opinion and the investigation of its relation to the opposing opinions.Exposition seeks to explain a situation through answering the questions of how and why.Cause and effect patterns (posing a problem, presenting a solution and supporting it through evidence and examples) are frequently used in this type of text.Instructional texts deal with planning someone's future behavior; hence, they tell the reader what to do.
Moreover, Harris (1990) classifies the topic sentences in paragraphs into five different types based on the way ideas are presented.The topic sentence may introduce the main topic of a paragraph, mention a fact or define the main idea of a paragraph, discuss the similarities or differences concerning an element, present a previous event, and refer to the evidence necessary for understanding some phenomenon.In effect, different types of topic sentences lead to the presentation of different methods of paragraph development, and different types of paragraphs (i.e., exemplification, comparison and contrast, classification, process, and causeand-effect).
Some studies have investigated the use of different cohesive devices in different types of genres (Abdul Rahman 2013, Alyousef & Alnasser 2015, Kudo 2009, Meisuo 2000, Rahayu & Cahyono 2015, Ramasawmy & Narainsamy 2004, Sayah & Fatemi 2013, Yang & Sun 2012).Using Halliday and Hasan's (1976) taxonomy of cohesive devices, Meisuo (2000) investigated the grammatical cohesive devices used in expository compositions written by Chinese EFL undergraduates.He concluded that conjunction was the dominant type of grammatical cohesive devices used followed by reference.Sayah and Fatemi's (2013) investigation of the application of conjunctions in narratives using story telling and story retelling by intermediate and advanced EFL learners showed that different types of conjunctions (i.e., additive, temporal, casual, and adversative) emerged as a function of learners' proficiency level, the content and the instruction.Adversative and casual conjunctions were overlooked by learners while additive conjunctions were applied by most EFL learners.As learners were not obliged to retell the stories using the same length as the original stories, different types of conjunctions were not used.Furthermore, when learners received instruction on different types of conjunctions through summarizing reading passages, they could rely more on conjunctions in their writing samples.
In addition, Abdul Rahman ( 2013), in his study of EFL learners' use of cohesive devices in writing descriptive texts, found that EFL learners overused specific types of cohesive devices (i.e., repetition and reference) and ignored other cohesive devices.Thus, the learners' written productions appeared to be non-cohesive and redundant.Kudo's (2009) investigation of grammatical cohesive devices in argumentative and narrative essays written by Japanese learners showed that more proficient learners used more conjunctions and reference in narrative essays than less proficient ones.However, in argumentative essays, more proficient learners used reference (specifically demonstratives).Thus, learners' proficiency level was considered a decisive factor in the use of cohesive devices.
In their investigation of conjunctive cohesion in EFL learners' narrative and expository writings based on Halliday and Hasan's (1976) taxonomy, Ramasawmy and Narainsamy (2004) showed that there were not any significant differences between high and low-rated writing samples in narrative compositions in terms of conjunctive cohesion.However, such differences were found in expository compositions.In other words, there was a positive relationship between conjunctive cohesion and the quality of writings.Furthermore, there was a relation between conjunctive cohesion and the genres of the writings.There were different distributions of every type of conjunctive (i.e., temporal, causal, adversative and additive) in narrative and expository writings.
According to Rahayu and Cahyono (2015), learners, writing exemplification essays, had a tendency to elaborate more and draw a conclusion rather than contrast ideas.This pattern is in connection with the nature of exemplification essays, which explicate topic with real examples.On the other hand, the higher frequency of elaborative markers in comparison and contrast essays indicated that the learners had a tendency to establish their concept of comparison and contrast by elaborating, contrasting, and inferring.In classification essays, however, learners tended to elaborate and make an inference from the elaboration and also provide the readers with a contrast regarding the related topic.Similarly, in process analysis essays, learners showed a preference for elucidating a particular process through using causal relations between segments and adding detailed information.Lastly, in cause-and-effect analysis essays, inferential markers were significantly used to create causal relations regarding a particular topic.A further finding of the study was that there were five common inferential markers: because, so, then, in conclusion, and therefore.Although because is less formal, it was frequently used to show a causal relation.Alyousef and Alnasser's (2015) investigation of cohesive devices in the Management Accounting texts comprising multimodal data showed that reference was the most frequent grammatical cohesive device.This observation was accounted for by asserting that the topic was maintained through writing about the same entities.However, the participants underused substitution, ellipsis and conjunction.Although ellipsis rarely occurred in the participants' written texts, substitution was never observed.Additionally, conjunctions were minimally used to express addition and causation.The underuse of these devices in the participants' texts was accounted for by some reasons, including their use of rhetorical organization dominant in their first language, their limited writing opportunities, and the educational system which relied heavily on memorization.
In general, results of studies show that, in comparison to the others, some cohesive devices are more frequently used in learners' productions.This can be attributed to some reasons.First, data collection procedure may be a determining factor: some cohesive devices are prevalently used in oral data while some are commonly used in written data.For instance, Yang and Sun (2012) noted that ellipsis and substitution were most frequently used in spoken language whereas they were rarely used in formal written discourse.Secondly, one might not have sufficient amount of knowledge about these devices to use them and in fact a lack of proficiency may play an important role.Thirdly, cross-linguistic differences may lead to different use of cohesive devices by native and nonnative speakers.Fourthly, if one is taught English by an inexperienced instructor with limited knowledge of coherence and cohesion, he is not able to convey the information coherently and accurately (Ghasemi 2013).For example, Hasannejad et al. (2012) found that learners could use substitution appropriately when they were efficiently taught how to use it.Similarly, Mahmoud (2014) asserts that EFL university learners do not have a serious problem in using conjunctions if they are taught through formfocused instruction and if they practice using them.In fact, as Mahmoud claims, both proficient and poor learners use most connectors appropriately and accurately because connectors, which form a small proportion of words, can be learned through appropriate instruction.Nevertheless, EFL learners might have problems regarding the appropriate use of connectives due to inter-lingual and intra-lingual transfer and/or the received instruction.
The above review reveals that although some studies investigated the use of cohesive devices in different types of genres, there are a limited number of studies investigating the use of cohesive devices in different paragraph types in general, and cohesive devices in different paragraph types written by native and non-native speakers, in particular.Furthermore, the investigation of cohesive devices was almost exclusively devoted to the examination of conjunctives in contrastive studies.Research in this area gains more significance in the EFL context of Iran when one considers the fact that earlier studies done in this context have evidenced inadequate or improper use of cohesive devices by Iranian EFL learners (e.g.Vahid Dastjerdi & Hayati Samian 2011, Zarepour 2016).In addition, to the best of the researchers' knowledge no study has examined the question whether raters' evaluation of paragraphs written by EFL learners may be affected by the cohesive devices used by the writers.The present study is an attempt to fill the above-mentioned gaps.

METHODOLOGY
The design of this study is mixed methods; the study employs both quantitative and qualitative approaches.In fact, following Halliday and Hasan's (1976) taxonomy, the grammatical cohesive devices were quantitatively investigated.Moreover, the perspectives of raters concerning the quality of the texts (and the cohesive devices) used by the writers were qualitatively explored.

PARTICIPANTS
Thirty under-graduate EFL learners, 13 males and 17 females, studying in an intact class in an Iranian state university, participated in this study.The learners were majoring in English Language and Literature.Their ages ranged from 19 to 24.They had already taken a writing course to enhance their English writing ability.The writing course oriented them to the use of writing conventions and standards including the general writing skills as well as the basic components and features of writing (i.e., the thesis statement, supporting evidence, organization, unity, coherence and cohesion).The course was taught by an English language instructor who was experienced in teaching writing to EFL learners.

MATERIALS
For the analysis of the grammatical cohesive markers of non-native Iranian speakers of English, 60 paragraphs (i.e., 30 paragraphs of cause-effect and 30 paragraphs of chronology) written by the 30 intermediate EFL learners were used.Furthermore, for the purpose of analyzing the grammatical cohesive markers of native speakers, 20 paragraphs (i.e., 10 paragraphs of cause-effect and 10 paragraphs of chronology) were chosen randomly from among the sample paragraphs written by English native speakers which were presented in the books teaching paragraph writing and development.

RATERS
Four raters participated in the present study.These raters are described in detail below.Rater 1 was a PhD.candidate at Shiraz University.He had learned English in an EFL context.He had taught General English courses in schools, English language institutes and universities for eight years.However, he had only taught writing for one year at the intermediate levels.
As for the rating experience, he was familiar with TOEFL and similar standard rating scales although he had not used them in practice.Furthermore, he was familiar with some rating scales prepared by English language institutes, and had used them for two years.
Rater 2 was also a PhD.candidate at Shiraz University.He had learned English in an EFL context, too.He had taught General English courses in English language institutes for six years.Furthermore, he had taught writing for three years at the advanced levels in the language institutes.As for rating experience, he was familiar with TOEFL and similar standard rating scales and had also used them for two years.
Similarly, Rater 3 was a PhD.candidate at Shiraz University and had learned English in an EFL context.She had also taught General English courses in English language institutes and university for nine years.However, she had not taught any writing courses.As for rating experience, she was not familiar with TOEFL or similar standard rating scales or even the rating scales prepared by English language institutes.
Finally, Rater 4 was also a PhD.candidate at Shiraz University and had learned English in an EFL context.She had taught General English courses in schools, English language institutes and universities for eight years.Furthermore, she had taught writing courses for two years at the advanced levels.As for rating experience, she had already been familiar with TOEFL and similar standard rating scales for four years, and had used them for two years.
In sum, the raters had more or less similar background.All of them, being Ph.D. candidates, had the same level of education, and were experienced in teaching English though they differed in terms of their rating experiences to some extent.However, they got familiar with TOEFL iBT rating scale before the study was conducted and then they started rating the EFL learners' written paragraphs.

RATING SCALE
TOEFL iBT holistic writing scale was used as a writing rubric for rating the learners' written paragraphs.This scale provided scores ranging from 0 to 5. The scale encompassed the following criteria for assessment: if the writing addressed the topic effectively; if it was wellorganized and well-developed; if it included clearly appropriate explanations, exemplifications and details; if it displayed unity, progression and coherence; if it displayed consistent facility in the use of language; and if it demonstrated syntactic variety, appropriate word choice and idiomaticity, and grammaticality.

DATA COLLECTION PROCEDURE
Sixty sample paragraphs were gathered from the 30 participants in the final exam session of their writing course.In fact, for the purpose of evaluation of their ability in writing English paragraphs, each participant was asked to write, among other things, two types of paragraphs, namely a paragraph of chronology and a paragraph of cause-effect.As for the paragraph of chronology, the learners were supposed to write a sufficiently interesting, exciting or unusual experience they had while studying in high school or university.As for the paragraph of cause-effect, they were supposed to give reasons for their agreement or disagreement with the claim that "the era of silver screen is coming to an end and people will eventually lose interest in going to the cinema".
Furthermore, in order to investigate the dominant categories of grammatical cohesive devices in native speakers' writing, 10 paragraphs of chronology and 10 paragraphs of causeeffect were selected randomly from among the paragraphs written by native speakers in different books on teaching writing.agreement indices were obtained to estimate intra-rater reliability in rating each of the subcategories of the grammatical cohesion.The indices turned out to be .98,.95,.93 and .95for reference, substitution, ellipsis and conjunction, respectively.Furthermore, to estimateinter-rater reliability, a PhD.candidate studying TEFL at Tehran University, who was already familiar with Halliday and Hasan's (1976) taxonomy of cohesive devices as a result of the necessary training, was asked to investigate grammatical cohesive devices in the same24 randomly selected paragraphs.Subsequently, agreement indices were calculated.The indices turned out to be .96,.89,.91 and .95for reference, substitution, ellipsis and conjunction, respectively.
To answer the first research question, the proportions of grammatical cohesive devices (i.e. the frequency of occurrence of each cohesive device divided by the number of words in a paragraph) were calculated.Then, the mean of the proportions in each group of paragraph types was estimated.To investigate if the type of paragraph (chronology vs. causeeffect) had any effects on the proportions of grammatical cohesive devices used in Iranian EFL learners' writing samples and those of native speakers, a two-way multivariate analysis of variance (MANOVA) was conducted.
In the second phase of the study, to investigate if the use of grammatical cohesive devices affected the raters' evaluation of the paragraphs, the four raters were asked to rate 12 randomly selected paragraphs written by the six learners (i.e., each of the six learners' paragraphs of chronology and cause-effect was rated by all the four raters).Shortly after that, the raters were asked to comment on their ratings and to explain why they had assigned a certain score to a writing sample.In other words, they were expected to talk about the criteria they considered when rating each sample so that their attention to cohesion and different cohesive devices used in different samples could be investigated.

RESULTS
Table 1 shows the frequency and the mean of proportions of grammatical cohesive devices used in paragraphs of chronology and cause-effect written by the EFL learners.It indicates that in their paragraphs of chronology, the EFL learners used reference (18.9%)more frequently than conjunction (4.85%), ellipsis (.2%) and substitution (.09%), respectively.Similarly, in their paragraphs of cause-effect, they used reference (11.23%) more frequently than conjunction (3.4%), ellipsis (.17%) and substitution (.07%), respectively.Table 2 shows the frequency and the mean of proportions of grammatical cohesive devices used in paragraphs of chronology and cause-effect written by native speakers.It indicates that in their paragraphs of chronology, native speakers used reference (12.9%)more frequently than conjunction (4%), ellipsis (0%) and substitution (0%), respectively.However, in their paragraphs of cause-effect, they used reference (10.53%)more frequently than conjunction (3.2%), substitution (.08%) and ellipsis (0%), respectively.As mentioned above, to investigate if the type of paragraph (chronology vs. causeeffect) had any effects on the proportions of grammatical cohesive devices used in Iranian EFL learners' writing samples and those of native speakers, a two-way multivariate analysis of variance (MANOVA) was done.Preliminary assumption testing was conducted to check for normality, linearity, univariate and multivariate outliers, homogeneity of variance, and multicollinearity.No serious violations were noted.The value of Wilks' Lambda and its associated significance level (sig.) are presented in Table 3.As the significance levels are less than the alpha level of .05,one can conclude that the groups differed in terms of language background (sig.= .008)and text type (sig.= .000).Moreover, the interaction effect was nonsignificant (sig.= .104).In other words, native and non-native speakers followed similar patterns in the use of cohesive devices.To investigate if language background and text type had different effects on all categories (i.e., reference, substitution, ellipsis, conjunction) or some of them, tests of between-subjects effects were conducted, the results of which are presented in Table 4.As it is evident in the table, language background affected the use of reference (sig.= .004);itshould be noted that although the reported significance level for ellipsis (sig=.057)and conjunction (sig=.067)are insignificant, they are close to 0.05.Moreover, text type affected the use of reference (sig.= .000),too.The estimated marginal means of the use of cohesive devices in each type of paragraph written by different groups are presented in Table 5.As for the differences in the use of reference in terms of language background, non-native speakers (Mean= 15.05) used it more than native speakers (Mean= 11.71).As far as the use of reference in different types of text sis concerned, paragraphs of chronology (Mean=15.90)contained more reference than paragraphs of cause-effect (Mean= 10.86).With respect to the second phase of the study, Table 6shows proportions of the grammatical cohesive devices used in paragraphs of chronology and cause-effect written by the six randomly selected EFL learners identified as A, B, C, D, E and F and Table 7 shows the raters' perceptions of the quality of the two types of paragraphs written by each of these learners.Cause-effect .12.12.106.106.14.23 As Table 7 shows, regarding the paragraphs written by Participant A, Rater 1 believed that the cause-effect paragraph had higher quality than the chronology paragraph.He focused on the accuracy, organization, cohesion and coherence of the paragraphs.Rater 2 believed that both paragraphs of cause-effect and chronology suffered from lack of grammatical, developmental and organizational features and detailed explanations to the same extent.Rater 3 believed that the participant's paragraph of chronology had higher quality than the paragraph of cause-effect.According to her, both paragraphs were short; they displayed unity, organization, and complex syntactic structures.However, the errors in paragraph of chronology did not lead to obscurity.Furthermore, the paragraph of cause-effect needed more explanation.Rater 4 suggested that both types of paragraph suffered from insufficient exemplification, details and organization.
Regarding the paragraphs written by Participant B, Rater 1 believed that the causeeffect paragraph had higher quality than the chronology one.It was more organized and enjoyed higher readability, cohesion and coherence.Similarly, according to Rater 2, the cause-effect paragraph had higher quality than the chronology one.The paragraph of chronology was not accurate and fluent.Furthermore, it contained non-related information.However, the paragraph of cause-effect was well-organized, well-explained, well-developed and intelligible.However, Rater 3 believed that the paragraph of chronology had higher quality than the paragraph of cause-effect.Although showing good unity and coherence, the paragraph of cause-effect displayed poor grammatical structures and inappropriate choice of words.Rater 4 considered both paragraphs equivalent in length, grammatical and lexical errors.The score assigned to the cause-effect paragraph is higher than that of chronology (4/3).The written paragraph of cause-effect has higher accuracy, organization, cohesion and coherence.
Both paragraphs receive the same score (1).Both paragraphs suffer from grammatical, developmental and organizational problems.They have no detailed explanations.
The score assigned to the paragraph of chronology is higher than that of cause-effect (3.25/3).Although both paragraphs are short, they display unity, organization, complex syntactic structures.There are a number of errors in both paragraphs.
But the errors in the paragraph of chronology do not lead to obscurity.Paragraph of cause-effect needs more explanation.
Both paragraphs receive the same score (2).They suffer from insufficient exemplification, details and organization.

B
The score assigned to the cause-effect paragraph is higher than that of chronology (4/2).The paragraph of causeeffect is more organized and enjoys higher readability, cohesion and coherence.
The score assigned to the cause-effect paragraph is higher than that of chronology (3/1).The paragraph of chronology is not accurate and fluent.It contains irrelevant information.
However, the paragraph of causeeffect is wellorganized, wellexplained and welldeveloped.Although it is not accurate, it is intelligible.
The score assigned to the paragraph of chronology is higher than that of cause-effect (4/3).The paragraph of causeeffect suffers from poor grammatical structures and inappropriate choice of words.However, it enjoys good unity and coherence.
Both paragraphs receive the same score (5).Although both are too short, they contain limited number of grammatical and lexical errors.

C
The score assigned to the cause-effect paragraph is higher than that of chronology (3.25/2.75).Both paragraphs have the same level of accuracy and organization.Paragraph of causeeffect enjoys more cohesion, coherence and readability.
The score assigned to the paragraph of chronology is higher than that of causeeffect (3/2).The paragraph of chronology enjoys more coherence and cohesion.The idea is also explained more.
The score assigned to the paragraph of cause-effect is higher than that of paragraph of chronology (4/3).The paragraph of causeeffect enjoys more coherence, unity and complex structures.
The score assigned to the paragraph of cause-effect is higher than that of chronology (3/2.5).It displays more unity and coherence.

D
The score assigned to the cause-effect paragraph is higher than that of chronology (5/4).Both paragraphs enjoy the same level of accuracy, readability, cohesion and coherence.The paragraph of causeeffect is more organized.
Both paragraphs receive the same score (4).Both paragraphs are fluent and accurate.They enjoy coherence, cohesion and appropriate explanation.
The score assigned to the paragraph of cause-effect is higher than that of paragraph of chronology (4/3).The paragraph of chronology is well organized, coherent and syntactically complex, but it needs more details and more elaboration.The paragraph of causeeffect enjoys more Both paragraphs receive the same score (5).They are well organized.They display unity, coherence and cohesion.
cohesion, supporting ideas, coherence, unity and organization.It contains grammatical errors.

E
The score assigned to the cause-effect paragraph is higher than that of chronology (4/3).The paragraph of causeeffect is more organized and enjoys more accuracy, readability, cohesion and coherence.Although it is shorter, it follows the mechanics of writing more accurately.
Both paragraphs receive the same score (1).Both paragraphs contain errors.They lack coherence, cohesion and sufficient explanation.
The score assigned to the paragraph of cause-effect is higher than that of paragraph of chronology (3/2).The paragraph of chronology lacks cohesion, appropriate organization and intelligibility.Although it is long, it contains redundant words and structures and ambiguity.The paragraph of causeeffect contains inappropriate words, grammatical structures and run-on sentences.However, the organization is good.
Both paragraphs receive the same score (2).They suffer from limited development of the topic, inadequate organization and lots of errors.

F
The score assigned to the cause-effect paragraph is higher than that of chronology (3.75/3.25).The paragraph of causeeffect is more accurate and readable.
Both paragraphs receive the same score (2).Both paragraphs lack proper organization, accuracy, and detailed explanation.
The score assigned to the paragraph of cause-effect is higher than that of paragraph of chronology (3/2).The paragraph of chronology gives explanation and exemplification.It lacks coherence and grammatical and lexical accuracy.Although paragraph of cause-effect contains obscure syntactic structures and is lessorganized, it is more coherent, and contains more explanation.
The score assigned to the paragraph of chronology is higher than that of cause-effect (4/3 Concerning the paragraphs written by Participant C, Rater 1 believed that the score devoted to the cause-effect paragraph was higher than that of the chronology paragraph.Although both paragraphs enjoyed the same level of accuracy and organization, the causeeffect paragraph displayed more cohesion, coherence and readability.Rater 2 believed that the paragraph of chronology had higher quality than the paragraph of cause-effect.It enjoyed more coherence and cohesion.Furthermore, the main idea was explained more.Rater 3 believed that the score assigned to the paragraph of cause-effect was higher than that of the paragraph of chronology because it displayed more coherence, unity and complex structures.Rater 4 suggested that the score she assigned to the paragraph of cause-effect was higher than that of the chronology paragraph because it displayed more unity and coherence.
As for the paragraphs written by Participant D, Rater 1 believed that the score he assigned to the cause-effect paragraph was higher than that of chronology paragraph.Although both paragraphs had the same level of accuracy, readability, cohesion and coherence, the paragraph of cause-effect was more organized.According to Rater 2, both paragraphs deserved the same scores.They were equally fluent and accurate, and they similarly enjoyed coherence, cohesion and appropriate explanation.However, Rater 3 believed that the paragraph of cause-effect had higher quality than the paragraph of chronology.Although the paragraph of chronology was well organized, coherent and syntactically complex, it needed more details and more elaboration.Furthermore, the paragraph of cause-effect enjoyed more cohesion, supporting ideas, coherence, unity and organization.It also contained fewer grammatical errors.According to Rater 4, both paragraphs deserved the same score because they were well organized and displayed unity, coherence and cohesion.
As for the paragraphs written by Participant E, Rater 1 believed that the score he assigned to the cause-effect paragraph was higher than that of the chronology paragraph because the paragraph of cause-effect was more organized and enjoyed more accuracy, readability, cohesion and coherence.Although it was shorter, it appropriately followed the mechanics of writing.According to Rater 2, both paragraphs received the same score due to the fact that both contained some errors.Furthermore, they lacked coherence, cohesion and sufficient explanation.Rater 3 believed that the paragraph of cause-effect enjoyed a higher quality than the paragraph of chronology.The paragraph of chronology lacked cohesion, appropriate organization and intelligibility.Although it was long, it contained redundant words and structures and ambiguity.The paragraph of cause-effect contained inappropriate words, grammatical structures and run-on sentences.However, it was well-organized.According to Rater 4, both paragraphs deserved the same score since they both suffered from limited development of topic, inadequate organization and errors.
As for the paragraphs written by Participant F, Rater 1 believed that the score he gave to the cause-effect paragraph was higher than that of the chronology paragraph because the former was more accurate and readable.According to Rater 2, the two paragraphs had similar properties.They lacked appropriate organization, accuracy, and detailed explanation.Rater 3 suggested that the score she assigned to the paragraph of cause-effect was higher than that of the paragraph of chronology because although the latter provided explanation and exemplification, it lacked coherence, as well as grammatical and lexical accuracy.According to her, although the paragraph of cause-effect contained obscure syntactic structures and was less organized, it was more coherent, and contained more explanation.Rater 4 believed that the score she assigned to the paragraph of chronology was higher than that of the cause-effect paragraph because it was more organized and it enjoyed more unity, coherence and cohesion.

DISCUSSION
As the results showed, regarding cohesive devices, native speakers and non-native speakers significantly differed only in terms of their use of reference.Native speakers used personal, demonstratives and comparatives more than non-native ones did.Just as Ghasemi (2013) argues, this can be accounted for by the writers' proficiency level and their level of linguistic knowledge.Furthermore, cross linguistic differences might be another reason for the emergence of this pattern.
As far as text type was concerned, paragraphs of cause-effect contained less reference than paragraphs of chronology.This lends support to what Rahayu and Cahyono (2015) noted: that conjunctions and inferential markers are more important than other features in conveying causal relations in paragraphs of cause-effect.
The qualitative phase of the study focused on whether the raters noticed cohesion in the writing samples while rating or not.The two paragraphs written by Participant A were a little different in terms of cohesion.The chronology paragraph included more cohesive devices as indicated in Table 6.From among the four raters, just one (Rater 1) considered the feature of cohesion in his rating.Rater 2, Rater 3 and Rater 4 totally ignored the feature.
Although Rater 1 considered cohesion in his rating, he incorrectly did it; he thought that the paragraph of cause-effect contained more cohesion (exactly the opposite of what Table 6 indicates).
As for the paragraphs written by Participant B, Table 6 indicates that the chronology paragraph enjoyed much higher cohesion.Rater 4 considered the two paragraphs similar, and Rater 1 and Rater2 considered the cause-effect paragraph higher in terms of quality.Only did Rater 3assign a higher score to the chronology paragraph.However, this rater did not pay attention to the feature of cohesion.Just one rater (Rater 1) considered the feature of cohesion in his rating, yet he did it incorrectly.He thought that the paragraph of cause-effect contained more cohesion (exactly the opposite of what Table 6 indicates).Raters 2, 3 and 4 were totally off-track and failed to notice the differences between the two paragraphs in terms of cohesion.
With regard to the paragraphs written by Participant C, Table 6 indicates that the chronology paragraph enjoyed higher cohesion.Three of the raters (raters 1, 3, 4) considered the quality of cause-effect paragraph higher than that of the chronology paragraph.Two raters (raters 1 and 2) considered cohesion a differentiating feature of the two paragraphs.Rater 2 correctly did it; according to him the chronology paragraph enjoyed more cohesion (as mentioned in Table 6).However, Rater 1 thought that the paragraph of cause-effect contained more cohesion (exactly the opposite of what Table 6

indicates).
When paragraphs written by Participant Dare considered, Table 6 indicates that the chronology paragraph enjoyed much higher cohesion.Two of the raters (raters 1, 3) believed that the paragraph of cause effect was of higher quality than that of chronology.Raters 2 and 4 assigned the same score to both paragraphs and considered them equivalent in terms of different features.All raters considered the feature of cohesion in their decision making.However, raters1, 2 and 4 did not consider it a distinctive feature while Rater 3 incorrectly did.She thought that the paragraph of cause-effect enjoyed more cohesion (exactly the opposite of what Table 6

indicates).
Regarding the paragraphs written by Participant E, Table 6 indicates that the chronology paragraph enjoyed higher cohesion.Two of the raters (raters 1, 3) believed that the paragraph of cause--effect was of higher quality than the chronology paragraph.Raters 2 and 4 considered the two paragraphs similar.Although three raters (raters 1, 2, 3) considered cohesion in their ratings, none of them correctly considered it a distinctive factor in the assessment of the paragraphs of chronology and cause-effect.
Finally, with respect to the paragraphs written by Participant F, Table 6 indicates that the cause-effect paragraph enjoyed much more cohesion.There was no consensus among the raters regarding the overall quality of the paragraphs.Raters 1 and 3 considered the quality of the paragraph of cause-effect higher in comparison to that of the chronology paragraph.Rater 2 believed that the two paragraphs were similar.Finally, Rater 4 admitted that the paragraph of chronology had higher quality.As for cohesion in the two paragraphs, just Rater 4did consider it.However, she did it incorrectly (exactly the opposite of what Table 6

indicates).
As it is evident in Table 7, some raters did not incorporate cohesion as a feature in their assessment of the paragraphs and considered the two paragraphs similar.Moreover, although some raters considered this feature distinctive when assessing paragraphs, they evaluated the cohesion of the paragraph exactly opposite to the way the paragraphs were characterized by the feature.In fact, a very limited number of raters noticed the feature in a small number of paragraphs.

CONCLUSION AND IMPLICATIONS
In the first phase of the study, it was found that native speakers and Iranian non-native speakers differed in the use of reference: native speakers used it more than Iranian nonnatives did.Furthermore, results concerning the text type showed that paragraphs of chronology contained more reference than paragraphs of cause-effect.In the qualitative phase of the study, as it was evident in Table 7, almost all of the raters considered coherence, organization and unity in their assessment of the quality of the paragraphs and their ratings.However, the feature of cohesion in the paragraphs was ignored by some raters.Furthermore, from among the raters who did take it into consideration, some were not consistent in their decisions on the quality of the paragraphs.For instance, Rater 2 considered cohesion in examining the paragraphs written by participants C, D and E, but not those written by participants A and B. Furthermore, from among the raters who considered cohesion, some did it incorrectly (i.e., their assessment of the mentioned feature was opposite to the actual feature of the paragraphs).To summarize, it might be concluded that although it seems that raters might be familiar with cohesion, they either do not use it in their ratings or they usually have a misconception of what cohesion is.
The study has some implications for material developers, teacher educators and teachers.Material developers need to pay more attention to the difference in the emerging patterns of reference as a cohesive device across different types of paragraphs.They can devote supplementary parts and activities in the books to introducing paragraph types in which learners may show inefficient use of reference when writing.Moreover, teacher educators should make teachers aware of the emerging difference in the use of reference in learners' performances across different types of paragraphs.In addition, in writing courses, teachers can devote the time required to present different types of paragraphs and cohesive devices in general and the dominant ones in each type in particular.

TABLE 1 .
Grammatical cohesive devices in paragraphs of chronology and cause-effect written by the EFL learners

TABLE 2 .
Grammatical cohesive devices in paragraphs of chronology and cause-effect written by native speakers

TABLE 3 .
The value of Wilks' Lambda in groups

TABLE 4 .
Tests of between-subjects effects

TABLE 5 .
Estimated marginal means

TABLE 6 .
Proportions of grammatical cohesion used in paragraphs of chronology and cause-effect written by the six EFL learners

TABLE 7 .
Raters' perceptions of the quality of the two types of paragraphs written by the six EFL learners